基于自适应高斯混合模型与静动态听觉特征融合的说话人识别

吴迪; 曹洁; 王进花

doi:10.3788/OPE.20132106.1598

您当前的位置：

首页 >

文章列表页 >

基于自适应高斯混合模型与静动态听觉特征融合的说话人识别

信息科学 | 更新时间：2020-08-12

- 基于自适应高斯混合模型与静动态听觉特征融合的说话人识别
- Speaker Recognition Based on Adapted Gaussian Mixture Model and Static and Dynamic Auditory Feature Fusion
- 光学精密工程 2013年21卷第6期页码：1598-1604
- 作者机构：
  
  兰州理工大学电气工程与信息工程学院
- 作者简介：
- 基金信息：
  
  国家自然基金(51375011)
- DOI：10.3788/OPE.20132106.1598
  中图分类号： TP391.41
- 收稿日期：2012-12-10，
  
  修回日期：2013-02-28，
  
  网络出版日期：2013-06-20，
  
  纸质出版日期：2013-06-15
- 稿件说明：
移动端阅览
吴迪曹洁王进花. 基于自适应高斯混合模型与静动态听觉特征融合的说话人识别[J]. 光学精密工程, 2013,21(6): 1598-1604

WU Di CAO Jie WANG Jin-hua. Speaker Recognition Based on Adapted Gaussian Mixture Model and Static and Dynamic Auditory Feature Fusion[J]. Editorial Office of Optics and Precision Engineering, 2013,21(6): 1598-1604
吴迪曹洁王进花. 基于自适应高斯混合模型与静动态听觉特征融合的说话人识别[J]. 光学精密工程, 2013,21(6): 1598-1604 DOI： 10.3788/OPE.20132106.1598.

WU Di CAO Jie WANG Jin-hua. Speaker Recognition Based on Adapted Gaussian Mixture Model and Static and Dynamic Auditory Feature Fusion[J]. Editorial Office of Optics and Precision Engineering, 2013,21(6): 1598-1604 DOI： 10.3788/OPE.20132106.1598.

摘要

对特征参数和高斯混合模型进行改进，提出了一种特征域和模型域混合补偿的方法用于解决说话人识别特征受噪声影响较大以及高斯混合模型随训练样本长度减小而性能下降的问题。通过模拟人耳听觉，给出了基于伽马通滤波器的伽马通滤波倒谱系数；考虑其只反映了语音的静态特征，提取了能够反映语音动态特征的伽马通滑动差分倒谱系数。基于因子分析技术，利用移动因子表示高斯混合模型的自适应过程，通过训练语料较充分的说话人模型中的均值向量补偿受训练语料长度影响较大的分量的均值向量。仿真实验表明：在纯净背景下，本文方法的识别率达到了98.46%；在不同噪声环境下，本文提出的混合补偿方法能有效提高说话人识别系统的性能。

Abstract

By optimizing the feature vectors and Gaussian Mixture Models(GMMs)

a hybrid compensation method in model and feature domains is proposed. With the method

the speaker recognition features effected by the noise and the declined performance of GMM with reducing length of the training data under different unexpected noise environments are improved. By emulating human auditory

Gammatone Filter Cepstral Coefficients(GFCC) is given out based on Gammatone Filter bank models. As the GFCC only reflects the static properties

the Gammatone Filter Shifted Delta Cepstral Coefficients(GFSDCC) is extracted based on Shifted Delta Cepstral. Then

the adaptive process for each GMM model with sufficient training data is transformed to the shift factor based on factor analysis. Furthermore

when the training data are insufficient

the coordinate of the shift factor is learned from the GMM mixtures of insensitive to the training data and then it is adapted to compensate other GMM mixtures. The experiment result shows that the recognition rate of the method proposed is 98.46% . The conclusion is that the performance of speaker recognition system is improved under several kinds of noise environments.

关键词

Keywords

references

KINNUNEN T, LI H Z. An overview of text-independent speaker recognition: from features to supervectors [J]. Speech Communication, 2010,52:12-40.[2]HAMID R,SEYYED A ,HOSSEIN B,et al.. A new representation for speech frame recognition based on redundant wavelet filter banks [J].Speech Communication, 2012, 54:256-271.[3]TYLER K P, STEPHANIE N,JOHN D,et al.. Human voice recognition depends on language ability [J]. Science, 2011,333:595.[4]PARVIN Z,SEYYED A. Robust speech recognition by extracting invariant features [J].Procedia - Social and Behavioral Sciences, 2012,32(3):230-237.[5]SHAO Y,JIN ZH ZH,WANG D L. An auditory based feature for robust speech recognition [C]. ICASSP,2009:4625-4628.[6]MAK B K W, LAI T C, TSANG I W, et al.. Maximum penalized likelihood kernel regression for fast adaptation [J]. IEEE Transactions on Audio, Speech and Language Processing, 2009, 17(7): 1372-1381.[7]翟优，曾峦，熊伟.基于不变特征描述符实现星点匹配[J].光学精密工程, 2012,20(11):2531-2539. ZHAI Y,ZENG L, XIONG W.Star matching based on invariant feature descriptor [J].　Opt. Precision Eng., 2012,20(11):2531-2539. （in Chinese）[8]DU J,HUO Q.A feature compensation approach using high-order vector taylor series approximation of an explicit distortion model for noisy speech recognition[J].IEEE Transactions on Adio, Speech, and Language Processing,2011,19(8):2285-2293.[9]JEONG Y. Speaker adaptation based on the multilinear decomposition of training speaker models [C]. Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing. Dallas, USA: IEEE, 2010:4870-4873.[10]HE Y J,HAN J.Gaussian specific compensation for channel distortion in speech recognition [J]. IEEE SIGNAL PROCESSING LETTERS, 2011, 18(10): 599-602.[11]OMID D,BIN M,ENG S,et al.. Discriminative feature extraction for speech recognition using continuous output codes [J]. Pattern Recognition Letters, 2012,33:1703-1709.[12]史思琦,石光明,李甫.基于轮廓特征多层描述和评价的部分遮挡目标匹配[J].光学精密工程,2012, 20(12):2804-2811.SHI S Q,SHI G M,LI F.Partially occluded object matching via multi-level description and evaluation of contour feature [J].Opt. Precision Eng., 2012,20(12):2804-2811.（in Chinese）[13]BALWANT A. SONKAMBLE,DOYE D D. A novel linear-polynomial kernel to construct support vector machines for speech recognition[J].Journal of Computer Science, 2011,7 (7): 991-996.[14]TOMAS P,PETER R. Real-time recognition of affective states from nonverbal features of speech and its application for public speaking skill analysis [J]. IEEE Transactions on Affetive Computing, 2011,2(2):66-78.[15]SANTHOSH K C, MOHANDAS V P. Robust features for multilingual acoustic modeling[J]. Int J Speech Technol ,2011, 14:147-155.

浏览量

648

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

利用高斯混合模型的多光谱图像模糊聚类分割

微型精密零件图像测量清晰度算法的选择与综合评价算法